This article (Haan and Dijkerman 2020) is an updated version of a previous review (Dijkerman and Haan 2007) (cited 322 times according to Scopus) where Dijkerman and De Haan proposed a somatosensory model inspired by the two-route model of visual system (Goodale and Milner 1992).

Based on:

• “Body Schema” and “Body Image” (Head and Holmes 1911)
• Multiple body representations (Berlucchi and Aglioti 1997)
• pathway from the APC via SII to the posterior insula for perceptual learning and recognition (Mishkin 1979)

1. early processing in APC, then higher association areas
2. recognition and perception from the APC via SII to the posterior insula
3. action from the APC via SII to the PPC
4. recognition and perception requires both insula and PPC
5. independency between the two streams

There is convincing evidence of crosstalk between pathways.

Dissociations between the two routes depend on the precise task conditions and there are now many examples of ‘perceptual information’ affecting action programming and vice versa. ( RHI impact on movement displacements and grasping actions )

1. the type of motor response given and whether this involves a predefined target;
2. visual versus proprioceptive targets when testing the effect of the illusion on motor responses;
3. active induction of the illusion involving movements made by the participant versus passive tactile stimulation

Somatosensory processing is not only for perception and the guidance of action.

1. haptic object recognition
2. body perception
3. body ownership
4. affective processing
5. action

Somatosensory processing ontributes to functions that are:

• inherently multimodal in nature
• involving cortical regions that process input from various modalities

examples:

• body perception relies on the body matrix
• haptic feature perception involves multimodal areas
• sensory processing about the hand during action is influenced by visual and somatosensory signals in the occipital and parietal cortex
• tactile memory involves multimodal processing in posterior parietal and frontal areas

Reviewing the Literature, the Authors observed that:

1- there are observation concerned structures that they had omitted in the first model 2- these structures are all highly interconnected

This suggests either one comprehensive or several overlapping networks.

Therefore, the Authors identified five main networks in a interactive, distributed network.

Cylinder block:

basic somatosensory processing unit dominated by somatosensory input.

1. SI in anterior parietal cortex (early processing)
2. SII
3. Thalamus

in SI simple features, such as stimulus location and duration, that lead to direction and velocity of a target moving on the body surface.

This Cylinder block provides the input for 5 networks:

• haptic object recognition
• body perception
• body ownership
• affective processing
• action representation

Recognizing objects needs:

1. learning
2. active exploration

• Memory of shape characteristics = left superior parietal lobe + right cuneus
• sensory-guided moter activity = left posterior parietal lobe

the insular cortex is NOT involved in object recognition or that its role is limited to the affective aspects of objects

in addition to posterior parietal areas, the frontal lobes also play a role in haptic recognition beyond programming exploratory movements

• monkeys: damaged SII impairments in tactile discrimination learning
• fMRI humans: anterior and superior parital cortex, SMA, dorsolateral and anterior cingulate cortex
• multimodal working memory (visuo-tactile amtrices): posterior parietal, dorsolateral prefrontal, anterior cingulate networks

prefrontal and parietal involvement in long-term tactile memory as well as a multimodal memory representation.

vision + somatosensory: activations in the medial and superior frontal gyrus and the superior parietal lobe including the intraparietal sulcus.

Imaging tactile features of an object accelerate reaction times to tactile but not auditory stimuli

Tactile imagery activation in: - SI - retrosplenial cortices - precuneus - prefrontal cortex \(\leftarrow\) top-down activation of a visual representation of a haptically explored object: haptic object recognition is multimodal!

Body perception depends on basic and higher-order representations.

• Spatial body information location of, distance of body parts, speed of touch sensation (\(\simeq\) body image?)
• Structural body information knowledge and awareness of the positions of body parts, knowing the left and right side (\(\simeq\) body schema?)

Structural and spatial representations are dissociable but they share underlying neural networks and show related functional representations.

• behavioural study suggests that there is a mental body representation structured in categorical body parts, with the joints as boundaries

• structural aspects of body representation affect early somatosensory processing (N80 from controlateral SI)

• finger agnosia and left-right disorientation: inferior parietal lobe damage, right anterior insula

Spatial aspects of body representations often studied with tactile distance estimation task (Spitoni et al. 2013).

Using the tactile distance estimation task it was observed the activation of:

• bilateral inferior and superior parietal areas
• right parieto-occipitotemporal junction.

Thus, parietal areas appear to be mainly responsible for processing spatial input.

Definition: the feeling that a body (or part of it) belongs exclusively to you

• three networks contribute in anosognosia for hemiplegia: the premotor loop, the limbic system and the ventral attentional network (Pacella et al. 2019)

• somatoparaphrenia mainly occurs after right hemispheric damage, usually posterior (temporoparietal) or insular but occasionally subcortically (Vallar and Ronchi 2009)

• in somatoparaphrenia are involved frontal (middle and inferior frontal gyrus) and subcortical (amygdala, hippocampus, thalamus, basal ganglia, white matter, internal capsule posterior limb) structures in addition to the postcentral gyrus (Moro et al. 2016)

• RHI: ventral premotor and posterior parietal areas, the EBA, the cerebellum, and the putamen to be involved + insula (Ehrsson, Spence, and Passingham 2004)

An important, but not the only, sensory channel for the positive affective aspect of touch is the c-tactile system

Unmyelinated c-fibres in hairy skin respond best to slow (1–10 cm/s) stroking with a soft brush or hand and at skin temperature (about 34°C), suggesting sensitivity for skin-on-skin contact (Rolls 2010).

These fibers directly aim to the Insular cortex.

The areas activated by affective touch are dissociable, but also overlapping the areas activated by disctiminatory tactile stimuli.

• Disciminative touch: SI, SII, the left lateral inferior premotor cortices, the inferior parietal cortex, the SMA, and bilateral angular gyri and medial prefrontal areas.

• Affective touch: bilateral network posterior and anterior insula, the postcentral primary and secondary somato-sensory cortex, the putamen, the thalamus, the frontal operculum, and the medial prefrontal cortex.

  1. feeling affective touch activates the posterior insula
  2. imagining affective touch activated the anterior insula
  3. observing affective touch activates the posterior insula (Morrison, Björnsdotter, and Olausson 2011)

This suggests that the anterior insula is particularly important for the affective interpretation of touch.

There are many other tactile stimuli that are socially relevant:

• holding hands
• hugging
• squeezing, etc

1. substantial overlap of the neural mechanisms underlying social and affective touch
2. both involve activation of somatosensory as well as social brain areas

Somatosensory (proprioception) information is used for:

1. starting position to plan the action
2. end position when we move towards a target location on the body

1. there is substantial crosstalk between the networks involved in action and perception.
2. a number of recent studies have specifically looked at the interplay between somatosensation and other sensory domains
3. dissociation in the haptic domain between grasping and reaching movements. Trajectories are different (Karl, Schneider, and Whishaw 2013)

This model is an interactive, distributed framework.

It shows functional specificity, but somatosensory processing in its broadest sense involves a network of cortical and subcortical areas that shows a more flexible organization than the first model previously assumed.

Questions

1. A central thesis here is the multimodal aspect of the networks that are informed by somatosensory information. The mechanisms are not well understood.

2. Affective somatosensory processing is not reliant on C-fibres only, and we need to look into the inter-action between these systems.

3. Time course of the activation of the different networks.